For this corpus I will make use of 100 popular Korean pop (K-Pop) songs and 100 western pop songs, both coming from the recent years. This corpus was chosen as I listen to a lot of songs from both genres and I really enjoy those. Therefor I would like to discover some more about the differences and similarities in songs from both these genres. This is also a relevant research as the K-Pop industry is becoming a larger worldwide presence and therefore this will also be interesting. As the genres as both of a pop category I would expect there to be a lot of similarities, but I still hope to find some differences.
As this corpus makes use of 100 popular songs from both these genres in the recent years I would argue that they represent their respective genre pretty well. One of the weaknesses of the corpus is that is makes use of a list not created by Spotify itself, therefore it may contain some personal preferences. An interesting mark for the corpus is that the K-Pop playlist contains some songs that are collaborations between Western and Korean artists.
Some typical track from the Western list are: “Shape of You, Ed Sheeran”, “Sugar, Maroon 5”, “Can’t feel my face, The Weeknd”. Some typical K-pop tracks are “Dynamite, BTS”, “I cant”t stop me, Twice”, “Kill This Love, BLACKPINK”. These are typical songs as these are from some of the biggest artist in their respective genres. Some interesting songs are: “More, KD/A” and “Kiss and Make Up,”Dua Lipa, BLACKPINK” as these are collaborated songs with Korean and western artists.
Links to the playlists:
Western
pop
Korean
pop
When looking at the differences between the energy levels of western pop songs and Korean pop songs you can certainly see a difference.
As is visible in the histogram the mean of the energy of western pop songs is around 0.7, while most of the Korean pop songs have an energy level higher than that. The K-pop songs have a mean energy around 0.85. This shows that K-pop songs have higher energy levels than most of the western songs.
MORE from K/DA is an interesting song as the group K/DA is a virtual group with their real life counterparts being Korean and Western artists. So the songs is made as a collaboration of western and Korean artist and therefore it would be interesting if its possible to see how their styles combine in the music. As is visible in the charts there is a clear tonal center around the C sharp. This seems to be more of a western style as it is more common in Korean music to have a some switches in the notes. Therefore while it is categorized as K-pop and it is sung by Korean artist the notes and cords still have a Western feel.
The korean group Dreamcatcher is one of my favourite groups and therefore I wanted to look at a self-similarity matrix from their recent song BEcause.In the graphs you can find the matrices for the chromatic features and the timbre of the songs. In chroma graph you can see that there are a lot of similarities, but also some differences at the start and at the end around 130. While listening to the song these can be related to the introduction and the bridge section at the end. At the start of the intro you can quietly hear the main chords, but the emphasis lies on a certain jumping sound that is not heard again later on in the song. The bridge section is totally different from the rest of the song as most music gets taken out of the song and there are only some subtle background sounds accompanied by different chords. The timbre graph also shows the same findings. The whole introduction is mostly accompanied by some drums and a piano, while the rest of the song is mostly accompanied by metallic guitars and drums. Also the bridge can be seen in the graph. Like said earlier most of the accompanying music fades away and all you hear are some high sung lyrics and some ticking and a piano.
Here are two tempograms of typical songs from the two genres. “I cant”t stop me” from Twice and “Shape of You” from Ed Sheeran. In both of the tempograms you can see a clear line that indicates the tempo of the song. The first is around 150bpm and the second around 90bpm. Looking at these visualisations you can see that the K-Pop song has a much higher bpm than the Wester pop song.
This plot visualizes the mean tempo of the songs in the two different playlist against their standard deviations. The duration of the songs are displayed in the sizes of the data points and the opacity indicates that how loud a song is. As is visible a lot of the songs from both playlists are in a fairly similar range in terms of their tempo standard deviations (around 0-1 bpm) as well as their mean tempo (around 100-150 bpm). This seems to be quite logical as pop music is forged from music that is appealing to most people. So they keep around the same foundation. What is also visible from the plot is that the duration and the loudness of most of the songs differs a lot as there are a lot of different sized data points with different opacity.
# A tibble: 2 x 3
class precision recall
<fct> <dbl> <dbl>
1 Korean Pop 0.829 0.861
2 Western Pop 0.854 0.82
For these visualization I trained and tested a Random Forest Classifier on my corpus. The results are pretty good as both precision and recall are around the 0.8.
According to the workflow the most important features for the classification are the first timbre component, loudness and the duration of the songs. This makes sense as earlier was also discovered that a lot of K-Pop songs have higher energy levels then the Western pop songs.
Visualizing these features in a graph makes it visible that these features indeed form a pretty good identifier for these genres. There is a decent possible separation line between the two genres based on these three alone. This would explain the good precision and recall as well.
After looking at both genres it is possible to come to some conclusions. There seem to be a certain amount of similarities between K-Pop and Western Pop, such as the tempo of the songs. There are however also some differences between them, such as the timbre, energy and duration. As the classifier shows it is almost perfectly possible to predict to which genre a song belongs according to these differences.Because of this I think that a lot of people are starting to like K-Pop as it still has some aspect of the music that they are used to in the West. The difference nonetheless still make it feel different and new. Which in turn makes it more enjoyable and interesting to hear.